1
Internet of Things in the 5G Era:
Enablers, Architecture and Business Models
Maria Rita Palattella, Member, IEEE, Mischa Dohler, Fellow Member, IEEE, Alfredo Grieco Senior
Member, IEEE, Gianluca Rizzo, Member, IEEE, Johan Torsner, Thomas Engel, Member, IEEE, and Latif Ladid
(Invited Paper)
Abstract—The IoT paradigm holds the promise to revolutionize
the way we live and work by means of a wealth of new
services, based on seamless interactions between a large amount
of heterogeneous devices. After decades of conceptual inception
of the IoT, in recent years a large variety of communication
technologies has gradually emerged, reflecting a large diversity
of application domains and of communication requirements. Such
heterogeneity and fragmentation of the connectivity landscape is
currently hampering the full realization of the IoT vision, by
posing several complex integration challenges. In this context,
the advent of 5G cellular systems, with the availability of a
connectivity technology which is at once truly ubiquitous, reliable,
scalable, and cost-efficient, is considered as a potentially key
driver for the yet-to emerge global IoT.
In the present paper, we analyze in detail the potential of 5G
technologies for the IoT, by considering both the technological
and standardization aspects. We review the present-day IoT
connectivity landscape, as well as the main 5G enablers for the
IoT. Last but not least, we illustrate the massive business shifts
that a tight link between IoT and 5G may cause in the operator
and vendors ecosystem.
Index Terms—Internet of Things, IoT, 5G, cellular, Low-Power
Wifi, Zigbee, Bluetooth Low Energy, Low Power Wide Area,
3GPP, Machine-Type Communications, MTC, Standardization
I. I NTRODUCTION
The idea of an Internet of Things as a network of smart
devices dates far back in the past [1], with the first applications
for automated inventory systems coming as early as 1983.
However, only from 1999 it took momentum, becoming part
of a shared vision for the future of Internet [2]. Today, the
growing pervasiveness and ubiquity, in almost any context, of
small and cheap computing devices, endowed with sensing
and communication capabilities, is paving the way to the
realization of the IoT vision.
A large variety of communication technologies has gradu-
ally emerged, reflecting a large diversity of application do-
mains and of communication requirements. Some of these
technologies are prevalent in a specific application domain,
M.R. Palattella, T. Engel, and L. Ladid are with the University of
Luxembourg, SnT, Luxembourg e-mail: {maria-rita.palattella, thomas.engel,
latif.ladid}@uni.lu.
M. Dohler is with King’s College London, UK email:
mischa.dohler@kcl.ac.uk.
L.A. Grieco is with Politecnico di Bari, Italy email: al-
fredo.grieco@poliba.it.
G. Rizzo is with the University of Applied Sciences Western Switzerland,
HES-SO Valais-Wallis, Switzerland email: gianluca.rizzo@hevs.ch.
J. Torsner is with Ericsson Research, Helsinki, Finland
email:johan.torsner@ericsson.com.
Manuscript received August, 2015; revised XXX, 2015.
such as Bluetooth Low Energy in Personal Area Networks
[3], and Zigbee in Home Automation systems [4]. Others, such
as WiFi, Low Power Wide Area Networks (LPWA) [5], and
cellular communications (such as 3GPP - 4G machine-type
communications, or MTC), have a much broader scope. In
addition, such landscape is constantly and rapidly evolving,
with new technologies being regularly proposed, and with
existing ones moving into new application domains.
A rough distinction is emerging between consumer IoT
(cIoT) and industrial IoT (iIoT) [6], with clear implications
on underlying technologies and business models. Consumer
IoT aims at improving the quality of people’s life by saving
time and money. It involves the interconnection of consumer
electronic devices, as well as of (virtually) anything belonging
to user environments such as homes, offices, and cities.
Conversely, industrial IoT focuses on the integration be-
tween Operational Technology (OT) and Information Technol-
ogy (IT) [7] and on how smart machines, networked sensors,
and data analytics can improve business-to-business services
across a wide variety of market sectors and activities, from
manufacturing to public services. It generally implies machine-
to-machine interactions, either for application monitoring (e.g.,
process monitoring in chemical production plants, vehicle fleet
tracking, among others), or as part of a self organized system,
with a distributed control which does not require human
intervention (i.e., autonomic industrial plants) [8].
Despite their evident differences, these two service domains
share some general communication requirements, such as
scalability, need for lean protocol stack implementations in
constrained devices, and friendliness to the IP ecosystem.
Nonetheless, the specific communication requirements of
iIoT and cIoT can be very different, in terms of reliability, QoS
(latency, throughput, etc), and privacy. cIoT communications
are typically machine-to-user, and usually in the form of client-
server interactions. In cIoT, desirable features of networked
things are low power consumption, ease of installation, inte-
gration and maintenance. Indeed, the quantified self paradigm
[9] which is currently unfolding with the advent of fitness
and health tracking systems, smart watches and sensor rich
smartphones requires a high power efficiency, in order to
enable long term monitoring by small, portable devices, as part
of a ”smart” environment or integrated in our daily wearings.
At the same time, such applications need to minimize
the risk of exposing such sensitive data as someone’s health
status or life habits. Increasing the number of nodes and of
exchanged information clearly multiplies the potential vulner-

abilities to the system to attacks and to privacy leaks [10].
Differently from cIoT, iIoT evolves from a large base of
systems employing machine to machine communications for
control process automation and/or monitoring. In such do-
mains, iIoT is the result of the integration, through the Internet,
of hardwired and often disconnected islands, usually based on
semi-proprietary protocols and architectures. Such integration
magnifies the potential of isolated industrial plants by aug-
menting their flexibility and manageability, and disclosing the
opportunity to deploy new services [8].
Many of iIoT communications, together with some of cIoT
communications, have often to satisfy stringent requirements
in terms of timeliness and reliability. Typically, the information
exchanged is critical for ensuring a correct and safe behavior
of the processes under control. Hence, the communication
network must be engineered in order to: (i) meet stringent
delay deadlines; (ii) be robust to packet losses; (iii) be safe
and resilient to damages, and more generally, strike the desired
balance between capital expenditure / operational expenditure
(CAPEX/OPEX) costs and system / service availability. 3G
and 4G cellular technologies, and especially 3GPP LTE [11],
are among the most appealing technologies in the modern
IoT connectivity landscape. They offer wide coverage, rela-
tively low deployment costs, high level of security, access to
dedicated spectrum, and simplicity of management. However,
being designed for optimized broadband communications, they
do not support efficiently MTC communications.
The advent of 5G communications1 represents a potentially
disruptive element in such a context. The increased data
rate, reduced end-to-end latency, and improved coverage with
respect to 4G hold the potential to cater for even the most
demanding of IoT applications in terms of communication
requirements. Its support for large amounts of devices enables
the vision of a truly global Internet of Things. In addition,
for its focus on the integration of heterogeneous access tech-
nologies, 5G may play the role of a unified interconnection
framework, facilitating a seamless connectivity of ”things”
with the Internet. The goal of this paper is to analyze in detail
the potential of 5G for the Internet of Things, considering both
the technological aspects and their implications on business
models and strategies.
The paper is structured as follows. Sec. II reviews the
main available IoT communication technologies, and their key
performance indicators. As MTC play a special role in such
context, in Sec. III we review MTC requirements and we
present the main standardization initiatives. Sec. IV describes
the main technical enablers of IoT in 5G. Sec. V describes
two MTC architectures, SmartM2M, and OneM2M. Sec. VI
presents the main business implications of 5G in the IoT
domain. Finally, Sec. VII concludes the paper.
II. M ODERN I OT C ONNECTIVITY L ANDSCAPE
Nowadays, the IoT landscape includes an extreme diver-
sity of available connectivity solutions which need first to
1 In this work, with the term 5G we refer to the solutions considered for and
specified by 3GPP from Release 15 onwards, including both LTE evolution
beyond Release 14 and a new 5G air interface. Eventually 5G is expected to
be defined by the technical solution(s) that fulfill the IMT 2020 requirements.
2
be harmonized across multiple industries, and then properly
combined together in order to meet the IoT technical Key
Performance Indicators (KPIs).
First forms of IoT connectivity can be dated back to the
80s, with the legacy Radio Frequency Identification (RFID)
technologies, and to the 90s, with the Wireless Sensor Net-
works (WSNs). Due to their attractive application scenarios,
both in business and consumer market, they gained a lot of
momentum. Therefore, for the first decade of the 21st century,
industrial alliances and Standards Developing Organizations
(SDOs)s put a lot of effort in developing standardized low
power IoT solutions. The first ones, available on the market,
were mainly proprietary solutions, such as WirelessHART,
and Z-Wave. They actually delayed the initial take off of the
IoT, due to interoperability issues, among different vendors.
Then, more generic connectivity technologies have been de-
veloped by SDOs, i.e., IEEE, ETSI, 3GPP, and IETF, easing
the interconnection and Internet-connection of constrained
devices. Bluetooth, and the IEEE802.15.4 standard [12] are
among the low power short range solutions available today,
which have played an important role in the IoT evolution.
Recently the IEEE802.15.4 physical (PHY) and medium ac-
cess control (MAC) layer have been complemented by an
IP-enabled IETF protocol stack. The IETF 6LoWPAN (today
6lo) [13] and IETF ROLL [14] WGs have played a key role
in facilitating the integration of low-power wireless networks
into the Internet, by proposing mainly distributed solutions
for address assignment and routing. At the same time, the
3rd Generation Partnership Project (3GPP) has been working
toward supporting M2M applications on 4G broadband mobile
networks, such as UMTS, and LTE, with the final aim of
embedding M2M communications in the 5G systems.
No one of these aforementioned technologies has emerged
as a market leader, mainly because of technology shortcoming,
and business model uncertainties. Now, the IoT connectiv-
ity field is at a turning point with many promising radio
technologies emerging as true M2M connectivity contenders:
Low-Power WiFi, Low-Power Wide Area (LPWA) networks
and several improvements for cellular M2M systems. These
solutions are very attractive for IoT deployments, being able
to fulfill availability and reliability requirements. With the final
aim of helping the understanding of this rich and variegated
context the reminder of this section overviews the modern IoT
connectivity landscape and characterizes in more details the
technologies which would potentially have a decisive impact
in enabling a global IoT in the upcoming future [15].
A. Zigbee
ZigBee is a low-cost, low-power, wireless mesh network
standard which has been widely applied in Wireless Sensors
Networks (WSNs)s, the first pioneering Industrial IoT apppli-
cations (e.g., for control and monitoring). ZigBee was initially
conceived in 1998, standardized in 2003, and finally revised in
2006. It builds on the IEEE802.15.4-2006 Physical (PHY) and
Medium Access Control (MAC) standard specifications [12].
From real deployments, realized so far, it emerged that the
current IEEE 802.15.4 PHY layer(s) suffice in terms of energy

efficiency. In fact, it is the actual hardware implementation
which dictates the exact current draws and thus the energy
needed to transmit a given information bit. However, many IoT
applications are expected to exchange only a few bits. Thus, it
will be advisable to look into a standardized PHY layer which
allows ultra low rate transmissions over very narrow frequency
bands, with the consequent advantage of having enormous link
budgets and thus significantly enhanced ranges [16].
From a MAC perspective, the IEEE802.15.4-2006 MAC
layer(s) did not suffice the needs of IoT applications. Its single-
channel nature makes it unreliable, especially in multi-hop
scenarios, where it incurs in an high level of interference
and fading. Moreover, it produces high energy consumption,
requiring router/forwarding nodes to be always on, regardless
of their actual traffic [16]. To overcome such limitations, the
IEEE802.15 Task Group 4e (TG4e) was created in 2008 to re-
design the existing IEEE802.15.4-2006 MAC standard and ob-
tain a low-power multi-hop MAC better suitable for emerging
embedded industrial applications. The IEEE802.15.4e standard
[17], published in 2012 as an amendment of the IEEE802.15.4-
2011 MAC protocol, defined three new MACs. Among these,
the Timeslotted Channel Hopping (TSCH) mode is the most
promising one, facilitating energy efficient multi-hop commu-
nications, while reducing fading and interference. The basic
concept on which TSCH builds on (i.e. the combination of
time synchronization and channel hopping) was introduced for
the first time by Dust Networks in 2006 in its proprietary Time
Synchronized Mesh Protocol (TSMP) [18]. The core ideas of
TSMP then made it into standards such as WirelessHART
(2007) [19] and ISA100.11a (2009) [20]. These standards
have targeted the industrial market, which requires ultra-high
reliability and ultra-low power. IEEE802.15.4e TSCH inherits
directly from these industrial standards, which are already de-
ployed as commercial products in ten of thousands of networks
in operation today. TSCH is thus a proven technology. One
important difference with existing industrial standards is that
IEEE802.15.4e TSCH focuses exclusively on the MAC layer.
This clean layering allows for TSCH to fit under an IPv6-
enabled upper protocol stack. The IETF 6TiSCH Working
Groups (WGs) [21], charted in 2012, has defined an open
standards-based Industrial IoT protocol stack which combines
IEEE lower layers (IEEE 802.15.4 PHY and IEEE 802.15.4e
TSCH), with IETF higher layers (6LoWPAN). Such protocol
stack designed for Industrial IoT networks, is able to fulfill
their stringent requirements, in terms of low latency, ultra low
jitter, and high reliability [7].
B. Bluetooth Low Energy (BLE)
In 2010 the Bluetooth Special Interest Group (SIG) pro-
posed BLE in the Bluetooth 4.0 specification [22], nowadays
updated to version 4.1. BLE is a smart low energy version of
Bluetooth, still designed for short-range communication (up
to 50 m), but, mainly suitable for low-power, control and
monitoring applications (e.g., automotive, entertainment, home
automation, etc.). BLE operates in the 2.4 GHz Industrial
Scientific Medical (ISM) band, and defines 40 channels, with
2 MHz channel spacing. To the final aim of achieving low
3
power usage, BLE uses (and thus, scans) only 3 advertising
channels, which are used for device discovery, connection set
up, and broadcast transmission. Their center frequencies have
been assigned to minimize interference with the IEEE802.11
channels 1, 6, and 11, widely used in several countries.
The other 37 data channels are dedicated to bidirectional
exchange of short bursts of data between connected devices.
BLE also sets up connections very quickly, which further
minimizes the radios on time. An adaptive frequency hopping
algorithm is used on top of the data channels, to reduce
sensitivity to interference, and multi-path fading [3]. BLE
has emerged in parallel with other low-power solutions, such
as ZigBee, 6LoWPAN, and Z-Wave, which were targeting
applications with multi-hop scenario. However, BLE currently
only supports a single-hop topology, namely piconet, with one
master device communicating with several slaves node, and a
broadcast group topology, with an advertiser node broadcast-
ing to several scanners. In 2015, the Bluetooth SIG announced
the formation of the Bluetooth Smart Mesh working group
to define the architecture for standardized mesh networking
for BLE. This will enable extended communication range
and simplify deployments of BLE networks for IoT. BLE is
destined to be a key enabling technology for some short-
range Internet of Things applications, such as in healthcare,
smart energy, and smart home domains [4]. Its potential
was recognized since its early birth, as shown by the interest
it gained quickly at IETF, where the 6LoWPAN Working
Group (today 6lo), developed a specification for allowing the
transmission of IPv6 packets over BLE [23]. BLE is expected
to become a de facto standard for short-range IoT services.
In fact, more than one billion smartphones are shipped every
year, all equipped with BLE interfaces. As a consequence,
this technology will become very soon the most commonplace
communication medium for consumer applications: from one
hand, a so broad market will let decrease the costs of BLE
hardware; on the other hand, motes equipped with the BLE
stack will have much chance to be used in all those scenarios
where user-to-machine interactions are needed (i.e., smart
living, health care, smart building, and so forth) [24].
C. Wifi and Low-Power Wifi (LP-Wifi)
The IEEE802.11 standard, better known as Wifi, was re-
leased in its first version in 1997, and designed without IoT in
mind. In fact, its final aim was to provide high throughput
to a limited number of devices (called stations), located
indoor, at a short distance between each other. Even though
nowadays widespread, Wifi has not been applied into M2M-
IoT use cases, due to its large energy consumption, fairly high
compared to other standards. For instance, Bluetooth, with
its shorter propagation range, (between 1 and 100 m), offer
a lower power consumption; while ZigBee have fairly long
range, but much lower data rate.
To overcome such energy-related limitation (which impacts
on the battery life of devices), duty cycling and hardware op-
timization have been adopted by the IEEE802.11 community,
achieving in this way, extremely energy efficient solutions.
Despite the reduced energy consumption, Wifi still suffers of

poor mobility and roaming support. In fact, it does not offer
any guaranteed QoS, and it is affected by high interference,
due to sharing the unlicensed 2.4 GHz band, together with
ZigBee, Bluetooth, and many others ISM band devices.
To reduce interference issues, the IEEE 802 LAN/MAN
Standards Committee (LMSC) has proposed the use of the sub
1GHz (S1G) license-exempt bands, which have better prop-
agation properties, especially in outdoor scenario, compared
to traditional WiFi 2.4 ad 5 GHz bands. It is possible to
increases the transmission range up to 1 km, while still using
the default transmission power of 200 mW. However, due to
the extremely scarce available spectrum, S1G does not allow
the use of wide bands (like those adopted by .11n, and .11ac,
> 20 MHz). Therefore, accurate design considerations, such
as new modulation and coding schemes were introduced in
the .11ac amendment.
In 2010, LMSC formed the IEEE802.11ah Task Group
(TGah), also called Low-Power Wifi, responsible of extending
the application area of Wifi networks, to meet the IoT require-
ments (large number of devices, large coverage range, and
energy constrains). In many IoT applications (e.g., smart grids,
industrial automation, environmental monitoring, healthcare,
fitness systems), an access point (AP) has to cover hundreds,
or even thousands of devices (sensors and actuators), which
periodically transmit short packets. One of the main challenges
for the adoption of legacy IEEE802.11 has been the limited
number of stations that can be simultaneously associated
with the same AP. The IEEE802.11ah standard overcomes
such shortcoming, by introducing a novel hierarchical method
which defines groups of stations, and allows support for large
number of devices [25].
IEEE802.11ah uses sub-1GHz frequency bands and the
PHY layer design is based on the IEEE 802.11ac standard.
To accommodate narrower channel bandwidths, the IEEE
802.11ah physical layer is obtained by down-clocking ten
times the IEEE 802.11ac physical layer. The TGah has put
a lot of effort into improving the IEEE802.11 MAC layer to
obtain power efficiency, and reduce the overhead due to (a)
short packet transmission, and (b) long time features of Wifi
PHY. Both issues implies a large amount of channel resources,
occupied by frame headers, interframe spaces, control and
management frames, and wasted for repeated channel access
procedure (following collisions). New short frame formats,
advanced channel access mechanisms, novel power manage-
ments mechanisms are among the solutions designed by TGah,
which allows low-cost connectivity with Wifi in unlicensed
bands. First performance studies [26] show that IEEE802.11ah
will support a large set of M2M scenarios, such as agriculture
monitoring, smart metering, industrial automation. It will be
able to provide a QoS level higher than currently provisioned
in mobile networks, and enable scalable and cost-effective
solutions.
D. Low Power Wide Area (LPWA)
The LPWA technology has recently emerged specifically fo-
cusing on low-end IoT applications which require low cost de-
vice, long battery life time, small amounts of data exchanged,
4
an area for which traditional cellular M2M systems have not
been optimized . The term LPWA which was introduced by
Machina Research to the market, stands for high reach, low
cost, low power Wide Area Networks [5]. Primarily designed
for M2M networking, it operates in unlicensed spectrum, and
it is currently available in many different proprietary solu-
tions (Amber Wireless, Coronis, Huawei’s CIoT, LoRa, M2M
Spectrum Networks, NWave, On-Ramp Wireless, Senaptic,
Sigfox, Weightless, among many others). While most of these
technologies have been present in the market for some time, it
is Sigfox with its Network Operator strategy that has recently
kick-started the LPWA market. Aware of its potential, the
LoRa Alliance, Sigfox, amd Weigthless are engaged in LPWA
standardization activities, aiming towards licensed spectrum,
and overcome interoperability issues which may delay its roll-
out. According to Machina Research, which has first forecast
the market opportunity for such technology in early 2013,
LPWA will allow to interconnect a large number of low-cost
devices (up to 60% by 2022, with 3 billions LPWA M2M
connections, by 2023), making the M2M solution business
profitable, and providing a platform to build a large IoT
business [5].
The key features of LPWA can be summarized as follow: (i)
wide area coverage ( up to some tenths of Km), (ii) low cost
communication, (ii) long battery life (up to 10 years from a
single AA battery), and (iv) low bandwidth communication.
The latter limits the LPWA range of applications to a set of
M2M use cases, characterized by low data rate, and infrequent
transmissions (few hundred bytes of data) [27].
LPWA is not intended to replace cellular connections, but
rather it will be complementary to existing cellular technolo-
gies. As discussed later in the paper 3GPP has also initiated
work to make cellular systems, such as GSM and LTE, more
suitable for low end MTC applications.
Despite its appealing and promising features, LPWA
presents some downsides, mainly due to the use of unlicensed
spectrum for long-range communication. Notably, the effective
radiated power (ERP) in that part of the license-exempt band
is heavily regulated in terms of allowed transmission powers
(after antenna gain), duty cycles and access mechanisms. Since
antennas at the basestation and at the IoT device have entirely
different gain capabilities, the link capabilities in up and
downlink are skewed with the uplink having a link budget
advantage of up to 19dB. While European regulation allows
for a boosted downlink power of 13dB, a difference of at least
6dB remains which means that truly symmetrical connectivity
cannot be guaranteed. This means that simple operations, like
sending an acknowledgement, cannot be executed seamlessly
as in 3GPP technologies. Consequently, only a limited set of
IoT applications can be supported through this technology.
Moreover, LPWA cannot fulfill the scalability requirements
of large-scale IoT deployments, due to an impeding spec-
trum congestion [28]. According to Cisco IBSG prediction
there will be 50 billion devices connected to the Internet by
2020. With such explosion of devices connected through IoT,
millions of devices may appear within the coverage area of
a single LPWA base station. Many of those will be using
other radio technologies that share the spectrum with LPWA,

such as LP-Wifi, Z-Wave, Zigbee, IEEE 802.15.4g, etc. All
these transmission will be perceived as interference by the
LPWA device, having low receiver sensitivity for long-range
communication.
Nevertheless, LPWA is expected to be a key enabler for
IoT deployments in early market rollouts and for limited IoT
applications.
E. 3GPP Cellular: MTC
Within the cellular context, the IoT connectivity solution
is referred to as machine-to-machine (M2M) and within the
3GPP standardization body it is referred to as machine-type
communications (MTC). The industry vision is to enable
connectivity between machines in an autonomous manner
where such a connectivity was traditionally facilitated by
means of wires. Even though the wired M2M market will
remain , scalability can only be achieved through an untethered
approach. Wireless MTC solutions have thus emerged which
offer viable business benefits but also exhibit shortcomings.
MTC is attractive when compared to wired solutions for
a variety of reasons, such as robustness against single point
of failures or ease of deployment. The main challenge for
wireless solutions however is related to the cost of the radio,
the power consumption but also the variety of M2M services.
Compared to the other technologies which were reviewed
in previous sections, cellular MTC is able to offer quality
of service (QoS) support, mobility and roaming support, as
well as billing, security and global coverage. Another area
where cellular MTC excels today is the ability to connect
the sensors/devices through a standardized API to the core
enterprise systems, all in real-time, scalable and secure.
Despite the convincing advantages, some serious challenges
remain to make MTC an underlying connectivity backbone for
the Internet of Things. These challenges affect the device and
networking levels, as well as viability of business models.
In the following Sec. III-IV, we describe all the current
initiatives in 3GPP, and the effort still needed for making MTC
a key enabler of the IoT ecosystem.
III. 3GPP MTC R EQUIREMENTS & S TANDARDS
One of the major differences between the 4G and 5G design
efforts is the support of a truly large number of devices so as
to enable the vision of a truly global Internet of Things. Given
the vast array of connectivity technologies developed over past
decades (presented in previous Section II), the emergence of a
cellular technology as a true contender may seem rather sur-
prising since power consumption is significant and both capital
expenditure (CAPEX) and operational expenditures (OPEX)
perceived to be very high. On the other hand, the undisputed
global coverage, the massive inter-operable ecosystem and the
ability to service critical data traffic give a lot of credibility
to this technology family.
To this end, in this section we review in more details the 4G
Generation, the MTC requirements and we explain respective
standardization initiatives.
5
A. MTC Generation Tradeoff
Considering the stringent requirements of M2M, the 2G
technology family, i.e. GSM and GPRS/EGPRS, is ideal as
power consumption and cost are low, coverage global and the
eco-system is developed; however, from an economic point of
view it is perceived to be much more viable to revise the
bands for next generation systems.
The 3G family, i.e. UMTS and HSPA, has a lower power
efficiency and higher modem cost than 2G. The capabilities
exceed the requirements of many low end IoT applications and
may therefore be a less preferable choice for such applications.
3G has however proven popular for e.g. automotive M2M
applications and other more demanding M2M applications due
to the wide range of data rates required.
4G technologies, i.e LTE and LTE-A, are interesting again
since the capabilities meet the requirements for very demand-
ing MTC applications; the air interface, OFDM(A), allows the
scaling of the bandwidth according to needs. Modem cost
in early LTE releases is however an issue and the coverage
is in some markets still patchy even if coverage increases
quickly on a global level. A further argument for use of
4G LTE for MTC applications is to benefit from improved
spectral efficiency and the bandwidth flexibility offered by 4G
systems and longevity of the technology as a future cellular
system. Uptake, due to a rather patchy 4G coverage, is only
catching up now and the challenge of adapting a 4G LTE
system (that was specifically designed for efficient broadband
communication) to also deliver and support MTC applications
remains.
5G may thus be a timely technology offering lower cost,
lower energy consumption and support for very large number
of devices. Indeed, these requirements are at the forefront of
the 5G MTC design and − if successful − will undoubtedly
be an integral part of the Internet of Things in years to come.
The requirements and standardization initiatives which lead
into these design developments are reviewed in the following
Section III.
B. MTC Technical Requirements
3GPP cellular systems were primarily designed for human
voice and data use, and less so for machine needs. A important
point to consider for all the requirements is backwards com-
patibility. This is because IoT devices, whether cIoT or iIoT,
will likely stay for a longer time in the field which impacts
technology migration, among others.
1) Need for ”Zero-Complexity”: One of the most stringent
requirements are those on lowering device complexity to vir-
tually zero. This will positively impact the cost of the devices,
since silicon costs are virtually absent. To this end, 3GPP study
items identified several features that are not required for MTC
devices and could reduce device complexity significantly.
Notably, it was proposed for LTE to limit device capability
to a single receive RF chain, restricting supported peak data
rates to the maximum required by IoT applications, reducing
supported data bandwidth and support of half duplex operation
as key to reduce device complexity, among a few others.
Standardization work is needed to ensure that maintaining

system performance with normal 3GPP devices with additional
scheduler restrictions to serve these low complexity devices is
achievable. Considering the timeline, 3GPP has closed some
of the complexity reduction specifications for LTE in Release
12; the remaining and new complexity items are dealt with in
Release 13 and subsequent releases.
2) Need for Long battery life time: A large fraction of
the IoT devices will be battery operated and may be located
in remote areas where changing or charging batteries may
not be possible or economically feasible. Miniaturization of
devices also imply that the physical size of batteries will
be smaller which means that the total available energy in a
battery may not increase even if battery technology evolves.
The communication module in IoT devices therefore needs to
be very energy efficient in order to enable battery life times of
decades. A battery life time of 10 years is already feasible for
infrequent data transmissions with both LPWA technologies
and in LTE Rel-12 ; the challenge for 5G may therefore be to
allow battery life time of more than one decade also for more
frequent data transmissions.
3) Need for Coverage Improvement: Many of the industrial
and even consumer IoT applications will require high levels of
coverage. Examples of such applications are smart metering,
factory automation with basement coverage, etc. Many of the
connectivity business models only work if and only if almost
all devices in the network can be reached. Due to the nature
of the wireless channel providing 100% coverage including
indoor locations for example in basements is very costly.
There is a need to reach also the last few percent of devices
in challenging locations without adding significantly to the
total cost of the complete solution. Increasing the number of
base stations is in theory a solution but comes at the additional
cost of site acquisition/rental, backhaul provisioning, among
others. A viable approach could be to improve coverage
levels in some critical application contexts without adding
significantly to the overall cost of the solution. To this end,
3GPP is stipulating low complexity and improved coverage
MTC devices to facilitate a scalable IoT uptake. Notably,
coverage improvement is achieved by repetition of information
with more details provided in 3GPP Release 12, Stage 3.
4) Need for MTC User Identification & Control: A large
part of the low-cost MTC devices will have a SIM integrated;
however, it is envisaged that − for scalability, configuration
and complexity reasons − some MTC devices will not contain
a SIM. In that case it is essential to be able to individually
regulate access using prior defined SIM profiles. In general,
the SIM card contains the IMSI of the subscriber with direct
link to the HLR; the latter includes details about subscribed
MTC services and feature profile. Operators are already able
to support customized MTC services based on the subscription
profile, such as optimal data packet size, optimal routing
with dedicated Access Point Name (APN) for MTC services,
etc. Through the IMSI, specific charging policy for MTC
subscription is provisioned by the operator and the operator
has complete control over the subscriber that is allowed in the
network. 3GPP is likely to define one or more new LTE UE
categories for MTC. This will be one of the means to identify
and isolate MTC devices if they are impacting performance of
6
the network and be able to restrict access for MTC devices.
One of the concerns operators share is restricting access to
roaming devices. Notably, the operator should be able to
identify such roaming MTC devices from MTC specific user
equipment (UE) category and be able to restrict access to the
devices if the operator does not wish to service those devices.
5) Service Exposure and Enablement Support: 3rd party
support to the 3GPP system is rather important from a scalabil-
ity point of view. Therefore, service exposure and enablement
support are instrumental for MTC to succeed as a connectivity
solution for the emerging IoT. Standardization work related
to M2M service enablement is on-going in standardization
organisations outside 3GPP (e.g. ETSI TC M2M and the
oneM2M Global Initiative, described in Section V). 3GPP’s
support for service exposure and enablement however allows
3GPP capabilities to be natively offered outside the 3GPP core.
For this to work, additional information (e.g. transmission
scheduling information or indications for small data, device
triggering, etc.), and new interfaces between the 3GPP Core
Network and application platforms will need to be provided.
Importantly in this context, to ensure privacy for consumer
IoT, exposed network information needs to be delinked from
private user/subscriber information.
C. 3GPP MTC Standardization
3GPP technologies and their continuous enhancements are
staged in different releases in 3GPP (see Figure 1. An impor-
tant aspect contributing to the success of 3GPP technologies
for cellular use is maintaining backward compatibility with
legacy releases and tight interworking between technologies
and efficient roaming support which is also key for M2M/MTC
applications that require support for mobility.
3GPP specification work is roughly grouped into RAN,
SA, GERAN and CT. Each of these groups is responsible
for defining functions, requirements and interfaces of 3GPP
systems. More specifically, RAN is responsible for the radio
access part of 3G, 4G and now 5G. GERAN is responsible
for radio access part of 2G and its evolution. SA has the re-
sponsibility for the overall architecture and service capability.
And CT is responsible for specification of terminal interfaces
and capabilities and the core network part of 3GPP systems.
3GPP features are phased into releases and the work may
be preceded by a study to ensure a rough industry consensus
and a better understanding of the problem at hand; lately,
all new features are preceded by a study phase. 3GPP has
adopted the notion of releases to ensure a stable platform
for implementation while facilitating introduction of new
features.
While the idea of cellular underpinning the connectivity
of the emerging IoT is rather new, actual standardisation
work started as early as 2005! Notably, 3GPP TSG SA1
(group defining services) started with a feasibility study to
conclude with a report during 2007 as captured in 3GPP
Technical Report (TR) 22.868. 3GPP Release 10 Technical
Specification (TS) 22.368 specifies the Machine-to-Machine
communications requirements. Then, refinement of require-
ments, logical analysis was captured in 3GPP TR 23.888

Fig. 1. 3GPP releases and timeline.
and protocol implementation were staged for release 11 and
captured in respective TS document of responsible working
groups. Furthermore, charging requirements were addressed by
SA5 to reuse existing 3GPP functions (e.g. session initiation
and control) to the extent possible. 3GPP architecture work
on MTC started. In Rel-10 and in Rel-12 SA2 worked on
efficient transmission of Small Data Transmissions and Low
Power Consumption UEs. Ever since, the amount of study
items and technical specifications for MTC have increased
steadily and are now a core ingredient of standardization work.
The standardization of a new 5G air interface is foreseen
by 3GPP to be divided in phases where the first phase, with
a finalization targeted during 2018, is focusing on early com-
mercial deployments and a subset of the 5G requirements. The
second phase of the standardization, targeted to be finalized
at the end of 2019, targets fulfillment of the full set of 5G
requirements.
D. 3GPP Security
The security framework used in 3GPP systems was orig-
inally developed for GSM to provide a basic connectivity
service for human to human communication. The security
features of GSM were encryption of the air interface to avoid
eavesdropping and strong authentication mechanism of the
users. The main security solution was kept for 3G and 4G,
but enhancements were done to enhance the security level
such as introducing state of the art encryption algorithms,
more elaborate key management systems, integrity protection
of signaling and mutual authentication. The 3GPP security
framework is based on the tamper resistant SIM card, which
holds the credentials of the subscriber. By using the credentials
of the SIM card and corresponding credentials stored in the
network, the device and the network mutually authenticate
7
each other. The authentication mechanism also produces keys,
which are then used for encryption and integrity protection of
the communication on the radio interface. Subscriber privacy
in 3GPP systems is considered by using randomly assigned
temporary identifiers to make tracking of devices and users
more difficult.
Recently 3GPP have worked on enhancements specifically
aimed at MTC applications. The new requirements emerged
due to characteristics of MTC such as reduced signaling and
even 10 years battery lifetime are being taken into account
in the security work. 3GPP is working on to enhance the
security level of GPRS in order to support the so called
GPRS-based cellular IoT system [29]. Another work is to
develop security solutions for 3GPP systems to support very
low complexity and low cost devices targeting 10 years battery
lifetime [30]. The challenges of the removable SIM card to
meet the requirements of MTC, like remotely changing the
subscription and fitting a SIM card into a tiny device, were
studied in 3GPP some years ago, but the standardization work
was started in GSMA and ETSI, and it still continuing under
the name of embedded SIM.
IV. 5G I OT E NABLERS
In order to enable the ubiquitous connectivity required for
many of the IoT applications, many more features and func-
tionalities will need to be added to the currently predominantly
broadband approach. This inherently leads to a strong hetero-
geneous networking (HetNet) paradigm with multiple types
of wireless access nodes (with different MAC/PHY, coverage,
backhaul connectivity, QoS design parameters, among others).
HetNets will offer the required seamless connectivity for the
emerging IoT through a complex set of mechanisms for coordi-
nation and management [31]–[35]. Evolved 4G and emerging
5G networks will thus be characterized by interoperability and

integration between multiple radio access networks, including
unlicensed frequencies. The aim of this section is to review
recently finished 4G, currently ongoing 4G-Evolution as well
as emerging 5G design efforts towards accommodating IoT in
such a heterogeneous networking setting.
A. 4G-Evolution Feature Enhancements
Some of the requirements outlined in Section III have
already been addressed by the 3GPP community, which con-
stitutes an important step towards IoT connectivity. Some of
the most important solutions are briefly summarized below.
1) Narrowband operation: Work is ongoing in 3GPP to
specify a narrowband version of LTE named narrowband IoT
(NB-IoT) especially targeting MTC applications with low data
rate and need for low module cost and long battery life time.
With a bandwidth of 180 kHz NB-IoT can be deployed in
a re-farmed GSM carrier offering an alternative use of GSM
spectrum. Alternatively it can be deployed in the guard bands
of LTE spectrum allocations or using part of an operators LTE
spectrum. The details of the air interface design are still to be
settled and the higher layer protocols starting from Layer 2
and upwards will be common between NB-IoT and LTE. By
making the solution similar to LTE the large eco-system for
LTE modules can be mobilized to secure availability of device
chipsets and a fast rollout. The performance requirements for
NB-IoT are similar as for the wideband LTE MTC solution,
the main benefit lies in that the narrowband operation leads to
flexibility in the deployment.
2) Low Cost & Enhanced Coverage: 3GPP developed
techniques enabling reduced UE complexity and improved
coverage compared to normal LTE user equipment. Details are
captured in 3GPP TR 36.888 [11]. Release 12 introduces a new
UE category with reduced peak rate (1Mbit/s), a restriction in
MIMO modes to a single receive antenna and half duplex
operation.
Release 13 aims at further reducing the complexity by
specifying a reduced bandwidth (1.4 MHz) and a lower UE
power class to facilitate single chip implementations with
integrated power amplifier instead of having it as an external
component.
The Bill of Materials of a Rel-12 and Rel-13 MTC op-
timized device has been estimated to be by 3GPP [11],
respectively, 50 percent and 75-80 percent lower than for the
early Rel-8 Category 1 UE.
Finally, coverage improvement through data repetition is
enabled; but this feature is only recommended for delay-
tolerant MTC devices.
3) Device Power Saving: The main source of the energy
consumption for MTC devices in pre-Release 12 devices is the
periodical listening for possible paging messages which needs
to be done at least once every 2.56s (the maximum cycle of
the discontinuous reception (DRX)), and performing various
link quality measurements. For typical MTC traffic patterns,
the energy required for transmitting messages is only a small
fraction of the total energy consumed. A mechanism to reduce
the power consumption for MTC devices was introduced in
3GPP Release 12, the Power Saving Mode (PSM). In PSM
8
the device can turn off all functionality requiring the device
to listen while in idle mode. Consequently the device does not
receive paging messages or perform link quality measurements
in PSM. Since the device remains attached to the network,
less signaling is required compared to the approach where the
device would be completely shut off when not transmitting.
The device can transmit up-link data at any time but is only
reachable in down-link when the device has been active in up-
link, which happens at configurable time instances or when the
device has transmitted up-link data.
4) Overload & Congestion Control at RAN/CN: MTC de-
vices cause signalling overload as they simultaneously recover
triggering registration and other procedures causing increased
signalling and core network overload when trying to respond
simultaneously. Specifically on network failure, there could
be a sudden surge in signalling as MTC UE associated with
the network reselect to alternate network (possibly roaming)
simultaneously to the not yet failed network. A Rel-10 study
item on Network Improvements for Machine-type Commu-
nications (NIMTC) studied core network aspects of overload
and congestion control. A Rel-11 study item on RAN im-
provements for MTC was started Q4, 2009. A Rel-11 umbrella
work item on system improvements to MTC started in May
2010; this work item also addresses RAN overload control for
UTRAN and E-UTRAN. In essence, a mechanism to prevent
and control such scenarios is performed by configuring MTC
UEs ”low priority access” indicator identifying the devices
as delay tolerant devices; this can be used by the network to
control various procedures. In UTRAN or E-UTRAN the RAN
(RNC for UTRAN and eNB for E-UTRAN) would reject the
connection request from the UE with an extended wait time
of up to 30 minutes when overload is indicated by the Core
Network to the RAN.
5) Other Enhancements: 3GPP Release 12 dealt with fur-
ther MTC enhancements which were not addressed in previous
releases. For example, the issues for optimisations of clusters
of MTC devices, MTC group addressing, and other group
features have been considered through fine-tuned broadcast
message protocols. Furthermore, enhancements to the roaming
of MTC devices have been specified and optimized by using
the low access priority indication provided by the UE; a
eNB steers UEs configured for low access priority to specific
MMEs. Currently, 3GPP Release 13 and future releases are
expected to enable further support for efficient delivery of
MTC services and mass deployments of MTC devices.
B. 4G-Evolution & 5G RAT Enablers
We now consider some pertinent radio access technologies
(RATs) in evolved 4G as well as 5G systems. We discuss
their relevance to Internet of Things (IoT) and Machine-to-
Machine (M2M) services. Given that coverage extension was
such an important design item, we start discussing the role of
relaying. We then discuss the relevance of the much-discussed
mm-wave and device-to-device (D2D) technologies.
1) Relaying for Increased Coverage: Relaying is a key
technology for 5G systems [36]. In its classic formulation, it
is meant for extending the communication range of a Base

Station (BS) and improve throughput by means of Relay
Stations (RSs). In 4G/5G systems, relaying techniques can also
include chains of relaying nodes and mobile relaying nodes in
order to improve coverage, support bandwidth hungry services,
and provide connectivity to M2M systems. RSs can help
reaching users that are either outside or inside the coverage
of the BS [37]. In the first case, the RS simply extends the
perimeter of a cell so that it is possible to reach also users
that are far from the BS. In the second case, the RS enables
multipath diversity techniques to magnify the connectivity
strength to users that are weakly served by the BS.
Four different transmission schemas can be used by RSs:
Amplify and Forward, Demodulate and Forward, Selective
Decode and Forward, and Buffer Aided, which are ordered
by increased latencies and reliability [38]. With Amplify and
Forward, the RS simply repeats an amplified copy of received
signals. With Demodulation and Forward, received signals are
demodulated (without decoding) by the RS, which also makes
a hard decision before modulating and forwarding the new
signals to the next hop. With Selective Decode and Forward,
received signals are fully decoded and checked. In case of
successful decoding, the signals are encoded and forwarded
to the next hop. With Buffer Aided approaches data are
temporarily stored by the RS in order to encode and forward
them as soon as the channel quality is sufficiently high. This
kind of scheme can only be used for services not sensitive
to delay. Another relevant aspect for RSs is the spectrum
used for access and backhaul links, which can be overlapped
(inband relaying) or disjoined (outband relaying) [39]. With
outband schemes, RS can enable full duplex communications;
In contrast, with inband mechanisms, interference between
access and backhaul links can arise, so that time division
multiplexing is required.
With reference to IoT systems, relaying technologies could
improve the scalability of network access operations to ensure
the required coverage extension. . In fact, in cells with a
very high density of IoT devices it is possible to let them
associate to different RSs so that the burden of network access
is distributed among many nodes. In other words, without
RSs, all IoT devices within one cell associate to the same
BS, which causes network overload. In addition, the adoption
of RSs can improve the fault tolerance of the communication
infrastructure because the BS is no longer a single point
of failure. Because relaying technologies strengthen the sig-
nal to noise ratio, the reliability and timeliness of message
exchanged with IoT services are improved and the support
to mission critical applications becomes easier. For group
based communications, multicast services on top of multi-hop
networks of RSs need to be defined. Another challenge (which
is common to 4G and 5G systems) is the coordination of the
communication activities between the RSs in order to optimize
the channel usage with a limited signaling overhead.
2) Millimeter Wave Technologies: As the demand for ca-
pacity in mobile broadband communications increases dramat-
ically every year, wireless carriers must be prepared to support
up to a thousand-fold increase in total mobile traffic by 2020,
requiring researchers to seek greater capacity and to find new
wireless spectrum beyond the 4G standard. Recent studies
9
suggest that mm-wave frequencies could be used to augment
the currently saturated 700 MHz to 2.6 GHz radio spectrum
bands for wireless communications. The combination of cost-
effective CMOS technology that can now operate well into the
mm-wave frequency bands, and high-gain, steerable antennas
at the mobile and base station, strengthens the viability of
mm-wave wireless communications. Further, mm-wave carrier
frequencies allow for larger bandwidth allocations, which
translates directly to higher data transfer rates.
Among the main issues which need to be tackled are the
high propagation losses at the mm-wave frequencies, which
call for high density of antennas, particularly in difficult
environments like city centers; effects such as reflections and
fading pose serious technical obstacles which need to be
addressed. The great capacity offered by mm-wave would
enable high-rate MTC applications, such as automatic video
surveillance cameras. Furthermore, the short range of mm-
wave could be interesting for D2D situations (see below).
Albeit seemingly an overdesign for largely low-rate IoT appli-
cations, mm-wave offers the interesting possibility to construct
very, very short over-the-air data packets. This, in turn, allows
an even more aggressive duty cycling of MTC devices and
thus possibly vital energy savings to extend the IoT devices’
lifetime. In fact, low power mm-wave interfaces would be used
in D2D communications, thus saving the energy required by
the cellular radio interface.
3) Device-to-Device Communications: Device to Device
(D2D) communications represents a turning point in cellular
systems. They entail the possibility that two devices can
exchange data without the involvement of the BS or with just a
partial aid from the BS [40]. In contrast to WiFi and Bluetooth
technologies, which provide D2D capabilities in the unlicensed
band, with D2D communications the Quality of Service (QoS)
is controllable because of the use of the licensed spectrum. A
new generation of scenarios and services in 5G systems can
hence be enabled, including device relaying, context-aware
services, mobile cloud computing, off load strategies, and
disaster recovery. Four different types of D2D communications
can be distinguished:
• Device relaying with operator controlled link establish-
ment: any device can broaden the coverage of the BS, by
acting as a relay node.
• Direct D2D communication with operator controlled link
establishment: any pair of network nodes can directly
interact due to a D2D link, which is set up under the
control of the operator.
• Device relaying with device controlled link establish-
ment: the endpoints of a data session are in charge of
setting up a relaying infrastructure made of one or more
relaying devices.
• Direct D2D communication with device controlled link
establishment: any pair of devices can exchange messages
thanks to a D2D link, which is established without any
operator control.
Enabling IoT communications in the licensed band is essen-
tial to strengthening the support to mission critical applications
and group communications. Besides these advantages of D2D
communications, security, trust, interference management, re-

source discovery, and pricing issues should be addressed to
capitalize the potential of this technology. These issues become
very challenging when the D2D link is set up without any
involvement of the BS. Moreover, new business models are
required to answer the ”pay for what” question. In fact, devices
that act as relays will deplete their own resources (as battery,
storage, communication, and processing) to assist theD2D
model. Cross-operator D2D capability is an open challenge
complicated by the fact that FDD spectrum bands are different
for different operators.
C. 4G-Evolution & 5G RAN Enablers
T The RATs, illustrated in Sec. IV-B, will need to be
supported by a suitable RAN. Some interesting propositions
have been made w.r.t. RAN which are briefly discussed here.
1) Decoupled Down/Uplinks: The traditional notion of a
cell is changing dramatically given the increasing degree of
heterogeneity. Rather than belonging to a specific cell, a device
would choose the most suitable connection from the many
possible of connections available. In such a setting, given
that transmission powers differ significantly between downlink
(DL) and uplink (UL), a wireless device that sees multiple
BSs may access the infrastructure in a way that it receives the
DL traffic from one BS and sends UL traffic through another
BS. This concept had recently been introduced and is referred
to as Downlink and Uplink Decoupling (DUDe) [41]–[43].
DUDe was shown to yield significant throughput gains and,
more importantly in the context of the IoT, orders of magnitude
improvements in reliability. The high-level system architecture
is shown in Figure 2 where some IoT devices in a given area
are connected in DL/UL to the macrocell, some are connected
in DL/UL to the smallcell; and some have a decoupled access.
Fig. 2. DUDe system model for UL/DL decoupling.
2) License Assisted Access: Recently, the 3GPP com-
menced a work item on License Assisted Access (LAA),
where licensed and unlicensed carriers are aggregated. LAA
uses licensed spectrum for control-related transmissions while
sending data over both licensed and license-exempt carriers.
Whilst mainly designed for high-capacity applications, the
approach could be beneficial in the context of a ever-increasing
amount of IoT devices with increasing data rate demands.
Notably, all non-critical IoT traffic could be transmitted via the
licence-exempt band whilst being controlled from the licensed
band.
10
3) Radio Access Network as a Service: The solution space
of 5G management architectures spans from centralized to
fully distributed ones. Based on the degree of centralization
different scalability, stability, and optimality targets can be
reached. To sustain flexible management approaches the Radio
Access Network as a Service (RANaaS) concept has been
developed [44]. With the RANaaS, Radio Access Network
(RAN) resources are virtualized and exposed through cloud
platforms. By using this approach, management function-
alities can be split between BSs and the cloud based on
the degree of centralization that is required in any specific
networking context. Of course, when BS functionalities are
migrated to the cloud some extra delay may be incurred
in the , execution of management operations, so that the
degree of centralization should be proportional to the capacity
of the backhaul. RANaaS enables : (i) dynamic wireless
capacity allocation in time and space varying 5G systems;
(ii) increased flexibility of management platforms due to a
technology independent state representation of hyper-dense 5G
deployments; (iii) new stores of management functionalities
provided by third parties (i.e., new business opportunities);
(iv) easy coexistence of multiple operators, sharing the same
physical infrastructure.
With reference to IoT systems, self -* capabilities, includ-
ing self-healing, self-configuration, self-protection, and self-
optimization, can improve the flexibility and extensibility of
the communication infrastructure due to modern orchestra-
tion of available radio access and IoT technologies [45]. In
critical applications, self-* functionalities could allocate extra
bandwidth to those IoT devices that detected some dangerous
events and hence need to communicate at the maximum speed
(and with the highest reliability and timeliness). Also, self-*
functionalities can ease the optimization of radio resources so
that a higher number of devices can be connected to the net-
work, which is essential for a mass scale IoT deployments. The
key challenge related to self -* capabilities is tuning degree of
centralization and to enable radio resource optimization also in
presence of coexisting operators that share the same physical
infrastructure.
D. 4G-Evolution & 5G Network Enablers
Finally, we discuss the emerging network enablers which
are pertinent to supporting the above IoT RAT/RAN enablers.
From several possible enablers, we shall focus on software
defined networking (SDN) and Network Function Virtualiza-
tion (NFV).
1) Software Defined Networking: The Software Defined
Networking (SDN) paradigm initially designed for wired
networks (e.g., data centers), has recently gained a lot of
interest into the wireless environment [46], and it is seen
as a key technology enabler for 5G networks [47], [48].
SDN separates the data plane (i.e., the traffic forwarding
between network devices, such as switches, routers, end hosts)
from the control plane (i.e., the decision making about the
routing of traffic flow - forwarding rule) [49]. SDN cen-
tralizes network control into a logical entity, namely SDN
controller, and allows programmability of the network, by

external applications. With its centralized view of the network
(topology, active flows, etc.), SDN provides dynamic, flexible,
and automated reconfiguration of the network. SDN will be
able to address flexibility and interoperability challenges of
future multi-vendor, multi-tenant 5G scenarios. In fact, with
SDN it will be possible to deploy a vendor-independent service
delivery platform, able to proactive respond to the changing
business, end-users and market needs. Therefore, SDN will
simplify network design, management and maintenance in
heterogeneous networked environments [50].
With the explosion of devices connected through IoT, tra-
ditional network architectures will not be able to manage
both the volume of devices, and the amount of data they
will be dumping into the network. There will be need to
efficient manage the load of traffic and the network resources
in the 5G era, to avoid possible collapse of the network,
and allow the coexistence of different services with different
Quality of Service (QoS) requirements [51]. SDN will be
among the technologies addressing such issues in future IoT
applications, as envisaged by the current emerging activities
at IETF (DetNet, 6TiSCH).
2) Network Function Virtualization: Network Function Vir-
tualization (NFV) is a complementary technology of SDN,
destined to impact future 5G networks. NFV aims to virtualize
a set of network functions, by deploying them into software
packages, which can be assembled and chained to create
the same services provided by legacy networks. The NFV
concept comes from the classical service whereby many
virtual machines running different operating systems, software
and processors, can be installed on the same server. By mov-
ing network functions from dedicated hardware into general
purpose computing/storage platforms (e.g. servers), NFV
technologies will allows to manage many heterogeneous IoT
devices. Moreover, by implementing the network functions
in software packages that can be deployed in virtualized
infrustructure, NFV offers scalability and large flexibility in
operating and managing mobile devices. With NFV it will be
possible to reduce both CAPEX and OPEX. Currently, the use
of NFV is under discussion in the context of virtualizing the
core network, and centralizing the base band processing within
Radio Access Networks (RAN) [52].
V. MTC A RCHITECTURE IN 5G
M2M architectures allow the different actors of an IoT
system to exchange data, check the availability of resources,
discover how to compose complex services, handle device
registration, and offer a standardized output to any vertical
application [53]. Currently, the main challenge with M2M
architectures is the vertical fragmentation of the IoT market;
according to the Global Standards Collaboration Machine-to-
Machine Task Force, more than 140 organizations are in-
volved in M2M standardization worldwide. In this perspective,
the reasons behind the proliferation of M2M architectures
come from long ago; the first ancestors of the todays’
M2M systems were the industrial fieldbuses, designed in the
seventies to support process control applications. At that time,
each production plant was adopting its own M2M technology
11
and, as a legacy, the same approach has been applied, in
the last twenty years, to define currently available M2M spec-
ifications. Nowadays, each vendor adopts its own protocols
and data formats so that interoperability remains an utopian
requirement, while vendor lock-in issues worsen the quality-
price tradeoff and hinder the diffusion of IoT technologies.
Recently, two noticeably international standardization projects
(i.e., ESTI SmartM2M and oneM2M) have been formulated
to resolve fragmentation issues in M2M systems, based on
a RESTful design [54]. Both of them target the definition of
an horizontal service layer that is able to embrace different
existing communication technologies and to include future
extensions to 5G systems.
A. ETSI SmartM2M
One of the most relevant attempts to resolve fragmentation
issues has been put in place by European Telecommunications
Standards Institute (ETSI), which defined a horizontal service
platform for M2M interoperability. The resulting SmartM2M
standard platform is based on a RESTful Service Capability
Layer (SCL) [55] and it is accessible through open interfaces.
Using this horizontal service layer it becomes possible to set
up IoT services in a technology independent way.
The different instances of the SCL can run on top of
devices, gateways, and network instances (see also Fig. 3)
to enable generic communication, reachability, addressing ,
remote entity management, security, history and data retention,
transaction management, , and interworking proxy.
The smartM2M architecture is made of two domains: the
Device/Gateway Domain and the Network Domain. The for-
mer includes devices and gateways. The latter represents an
abstract system that enables all the services entailed by the
smartM2M architectures by leveraging the resources available
at the lower domain.
An M2M Device can run M2M Applications using a local
instance of the SCL. In this case, it is connected directly to the
Network Domain via the Access network and it may provide
services to other devices. It can also be connected to the
Network Domain via an M2M Gateway. The M2M Network
provides connectivity between M2M Devices and M2M Gate-
ways. An M2M Gateway also runs M2M Applications using
a local instance of the SCL and acts as a proxy between M2M
Devices and the Network Domain and may provide service to
other devices [53].
ETSI M2M adopted a RESTful architecture style, thus an
SCL contains a resource tree where the information is kept.
The resource tree structure that resides on an SCL as well
as the procedure for handling the resources have been also
standardized. A Resource is a uniquely addressable entity in
the RESTful vocabulary. Each resource has a representation
that may be transferred and manipulated with the Create,
Retrieve, Update, and Delete verbs. A resource shall be ad-
dressed using a Universal Resource Identifier (URI). Operation
on resource among Applications and SCLs, and between SCL
instances are supported by means of Methods that constitute
the communication on the several interfaces of the SmartM2M
architecture. Each method conveys a set of information defined
as Method Attributes.

An SCL resources tree (see also Fig. 4) includes different
kinds of resources as follows: sclBase, scls, scl, applications,
application, and henceforth. The sclBase resource describes
the hosting SCL, and is the root for all other resources
within the hosting SCL. The scl resource stores information
related to distant SCLs, residing on other machines, after
successful mutual authentication. The application resource
stores information about the application after a successful
registration on the hosting SCL. The container resource acts
as a mediator for data buffering to enable data exchange
between applications and SCLs. The contentInstance resource
represents a data instance in the container. The accessRight
resource manages permissions and permissions holders to
limit and protect the access to the resource tree structure.
The group resource enhances resources tree operations by
adding the grouping feature. For instance, a group resource
could be used to write the same content to a group of M2M
resources. The registration resource allows subscribers to
receive asynchronous notification when an event happens such
as the reception of new sensor event or the creation, update,
or delete of a resource. The announced resource contains a
partial representation of a resource in a remote SCL to simplify
discovery request on distributed SCLs. The discovery resource
acts as a search engine for resources. The collection resource,
groups common resources together.
Fig. 3. SmartM2M high level architecture.
With reference to security issues, the standard also defines
an M2M security framework, encompassing authentication,
key agreement and establishment, M2M service bootstrap, and
M2M service connection procedures, grounded on a clearly
defined key hierarchy of the M2M node [56].
The SmartM2M architecture is very flexible and extensible,
but, as pointed out in [57], it may suffer scalability issues
because all transactions are mediated by the M2M Network,
which can easily become a single point of failure or a
bottleneck.
B. From SmartM2M to oneM2M
With similar objectives but with a broader partnership, the
oneM2M Global Initiative has been recently chartered as an in-
12
ternational project. oneM2M also targets the definition a hor-
izontal service layer that interconnects heterogeneous M2M
hardware and software components on a global scale. oneM2M
has been kicked off by seven telecom standards organizations:
Association of Radio Industries and Businesses (ARIB) and
Telecommunication Technology Committee (TTC), Japan; the
Alliance for Telecommunications Industry Solutions (ATIS)
and Telecommunications Industry Association (TIA), United
States; the China Communications Standards Association
(CCSA), China; the European Telecommunications Standards
Institute (ETSI), Europe; and the Telecommunications Tech-
nology Association (TTA), Korea. These organizations in-
volve around 270 companies that are actively contributing to
oneM2M [58].
Fig. 4. ETSI URI resource tree structure (example).
From an architectural point of view, both oneM2M and
ETSI SmartM2M adopt a RESTful design, name resources
through a hierarchical name space, and are grounded on the
concept of horizontal service layer. In contrast to SmartM2M,
oneM2M relaxes scalability restraints by adopting an hier-
archical organization of the different actors in the system,
so that a logical tree of nodes is obtained, which include
application dedicated node (ADN), application service node
(ASN), middle node (MN), and infrastructure node (IN).
Nodes consist of at least one common services entity (CSE)
or one application entity (AE). A CSE is a logical entity
that is instantiated in an M2M node and comprises a set of
service functions called common services functions (CSFs).
CSFs can be used by applications and other CSEs. An AE is
a logical entity that provides application logic, such as remote
monitoring functionalities, for end-to-end M2M solutions.
The standard also defines a security architecture articulated
over three layers: security functions, security environment ab-
straction, and secure environment. Security functions include:
identification, authentication, authorization, security associa-
tion, sensitive data handling and security administration. Se-
curity environment abstraction offers many security primitives,
such as key derivation, data encryption/decryption, signa-

ture generation/verification, and security credential read/write
from/to the secure environments. This layer is not specified
in the initial release but is expected to be considered in
future ones. The security environment layer contains one or
multiple secure environments that provide different security
services related to sensitive data storage and sensitive function
execution [59].
VI. F LIP OF B USINESS M ODELS
The Internet has undergone a massive transformation from
being infrastructure-driven (Ethernet cables, routers, comput-
ers, etc.) to business-driven (Facebook, eBay, Google). The
Internet of Things is undergoing a similar transformation.
Today, we expend significant efforts to design sensors, con-
nectivity radios, gateways and basestations. In a few years’
time, efforts will hopefully be on business opportunities, e.g.
a ”Google of Things”; business opportunities here, however,
are strongly dependent on regulation around privacy and trust
in the technology. In this section, we therefore discuss this
transformation and the role the telecommunications ecosystem
will likely play.
A. Demand-Side Problem in IoT
The number of connected ”things” in the Internet of Things
is not meeting the predictions made a few years back. In
general, the IoT rollout lags behind in terms of what market
research has predicted. This is really surprising to many
since there is no doubt that instrumenting the planet with
IoT capabilities would yield significant operational savings
and/or financial gains. To understand the underlying market
dynamics, one has to understand that for new technologies
to succeed, three things need to come together: i) supply of
the technology itself; ii) proven business models which link
supply to demand; and iii) a strong market demand.
From previous sections of this paper, we can see that there
is a large supply of connectivity technologies and standards
available today. These have been tested and many of them are
used successfully in various deployment around the world.
Claims that there is a lack of technologies or standards to
unlock the IoT market are largely unfounded.
From a business modeling point of view, numerous models
are available today and some of them have been successfully
tested in real commercial deployments. For example, in the
smart city market, the city hall could use smart parking sen-
sors, smart garbage bin sensors and/or smart street- lighting
sensors. The smart parking sensors are not only able to guide
drivers to vacant parking spots (and thereby reducing driving
time, pollution, etc) but also correlate the occupancy data with
the payment data; the latter allows infringements to be spotted
more efficiently and thus improve the city’s financial income
from parking. The smart bin sensors are able to detect when
exactly the bin needs to be emptied, thereby improving pick-up
schedules and saving money to the city hall. The smart street-
lighting sensors are able to regulate the usage of the lamps
according to ambient light conditions, as well as movement in
the street (i.e. if nobody passes at 3am in the night, they switch
13
off); this yields an estimated saving of 30% in the electricity
bill in cities.
Why, given the supply of technology and strong business
models, is the IoT not taking off as quickly as we had hoped?
The reason is because market demand remains consistently
low. This is entirely normal with new technologies and mar-
kets. For instance, the Internet took more than a decade
to gain widespread use: people were doing accounting and
shopping for years without the Internet, to change that habit
took time. In the IoT smart city context: the city was manually
measuring air pollution for many years − why would they start
using autonomous sensors now? That is arguably the biggest
challenge for the IoT today, i.e. create a genuine demand
among industries and consumers. Once that demand is created
and procurement as well as supply chains adapted, the IoT will
take off exponentially, just as the Internet has around 2000.
B. IoT Return-of-Investment
To support the strength of today’s IoT business models,
let us examine the typical return of investment (ROI) metrics
used. The ROI is a major driver in the adoption of any
technology, as it indicates the ability to return the initial
financial investment. There are three major ROI arguments
to use IoT technologies:
1) Real-Time Instrumentation ROI: A study by General
Electric has identified the enormous efficiency benefits
stemming from real-time instrumentation by means of
industrial IoT technologies [60]. The study has looked
at verticals like transportation, health, manufacturing,
etc. While the study has not considered the cost of
the instrumentation, the strong ROI drive has become
evident.
2) Big Data Value ROI: Not so well quantified as of 2015,
there is understood to be an enormous value in cross-
correlating data from different verticals to give unique
insights which would not be evident on their own.
An example can be found in smart city transportation
where IoT data from traffic is cross-correlated with
weather and sports data, thus allowing to define viable
traffic management strategies on days where congestion
is likely. That is probably the most important ROI and
signals the true shift from an infrastructure to a data-
driven ecosystem.
3) Wireless ROI: A fairly straightforward but nonetheless
important drive in rolling out wireless IoT is the fact
that getting rid of cables allows achieving substantial
CAPEX and OPEX gains. While electronics and sensors
become cheaper over time, human labor and cable
costs (due to copper) increase. Going wireless saves
cable costs (CAPEX), installation efforts (CAPEX) and
system maintenance (OPEX).
Above ROI argumentations highlight that there is not only
utility in using IoT technologies but there are clear financial
returns. Once demand has picked up, the market up-take will
almost certainly be guaranteed.

C. Zero-Capex IoT Business Models
In the context of smart cities, many municipalities as well
as governments are striving to deliver a higher quality of
public services; however, they are usually challenged by the
lack of funding and poor financial capabilities [61]. So far the
most common means for a city to become smart is to receive
governmental funds; for instance, Glasgow (UK) recently won
a governmental grant of 24 M £ to become the countrys first
smart city. Even though these awards could be perceived as
a good start for smart city initiatives, such awards are not
self-sustaining of deploying IoT technologies.
Given IoT business models are generally solid (and ac-
cepted), the demand up-take can be accelerated by covering
the often large CAPEX costs by the operational gains. An
ideal scenario is where all up-front CAPEX costs are covered
by a bank through a loan which is being paid back over time
through the financial returns or savings. This approach is is
just beginning in the context of the IoT and smart cities, but
early results are available from [62].
Notably, [62] observes that most IoT business models have a
given CAPEX cost c and an operational income which scales
roughly linearly over time with gradient a. The operational
income is composed of the revenues/savings from using the
IoT technology, diminished by the OPEX costs (typically in
the order of 10% of the CAPEX cost per annum). Three
different models are investigated, i.e. i) the quickest rollout by
deploying all IoT technologies at once; ii) the lowest CAPEX
rollout by deploying the technology with the lowest c first and
then deploy the second-lowest c the moment it can be paid by
the first technology due to making financial gains; and iii)
deploy in such an order that risk to investors is minimized by
understanding the different growth estimation errors.
D. 3GPP’s Universal Control Capabilities?
Let us assume there is a massive demand for IoT by 2020
and the rollouts are increasing exponentially. Three operational
models are possible:
1) The ”Bluetooth Model”, i.e. where the consumer and
industry purchase the IoT equipment and handle con-
nectivity themselves. As explained in Section II, the
disadvantage is that many critical IoT applications will
probably not work well.
2) The ”Wifi Model”, i.e. local connectivity is provided and
largely managed through an IoT operator which bills
the customer on a monthly basis. While the operational
pain is taken out for the customer, the IoT operator is
not able to guarantee delivery of critical IoT data since
license-exempt spectrum is used.
3) The ”Cellular Operator Model”, i.e. truly global connec-
tivity is provided using a cellular IoT provider (which
does not need to be the traditional operator). While
a few technical challenges still need to be solved, the
advantage is obvious in that reliable and accountable
connectivity can be provided globally which is an im-
portant value add to both consumers and business relying
on IoT technologies.
14
The third ”Cellular Operator Model” opens up other inter-
esting operational, in addition to those already discussed above
in this paper. Notably, if/once 3GPP opens up to other IoT
connectivity technologies (see below), then it could act as a
truly global IoT control engine. Using 3GPP’s authentication,
security, billing and SLA capabilities (to mention a few), it
could provide all these services as an umbrella while the
underlying technologies can be 3GPP-based or non-3GPP-
based. Technology enablers, such as license-assisted access
(LAA), local IP access (LIPA), capillary networking architec-
tures (as developed in ETSI M2M), among many others, are
available already today. In addition, the recent proposition by
Vodafone to fast-track Neul2 -like technologies in 3GPP will
likely accelerate such a development.
From a business point of view, this means that 3GPP would
slowly shift from providing data pipes to rather controlling
an ensemble of data pipes. That would signal an important
shift for operators from business-to-consumer (B2C) driven
business to rather business-to-business (B2B) driven one. In
principle, this is an important opportunity to scale sales as
cellular servicing is extended to any available wireless system.
E. Shake-Up in Cellular Value Chain?
From above, it is obvious that the IoT could enable a
true transformation in the cellular ecosystem. While the per-
bit value of IoT is rather low, the value due to a holistic
orchestration and big data exploitation is enormous. Let’s
briefly examine the possible shifts in the ecosystem for both
operators and vendors.
Operators will add an important value due to global or-
chestration, relying on roaming and other agreements already
struck in the ecosystem. This saves a lot of operational hassle
and it difficult to reproduce by e.g. the Wifi community. Fur-
thermore, a lot of the IoT deployments will be critical which
requires e.g. application/content-hosting cloud technologies to
be placed right at the edge, i.e. into the operators networks. In
addition, net-neutrality regulations may apply less stringently
to machines and − through SDN − operators have all the
technical means to reserve special data pipes for IoT traffic
and thus giving an operator-run IoT deployment a cutting-
edge over other deployment approaches. For all this, operators
will be able to charge premium rates to an extent which other
communities are not able to do which is a fundamental shift
on how this market will run.
Vendors, on the other hand, will likely capitalize on the
unique characteristics of industrial IoT deployments. Notably,
that market will be heavily B2B-driven and less B2C-driven.
The latter is well-understood by operators while the former
is much better understood by vendors. We will thus likely see
many B2B alliances struck between a vendor and an industrial
IoT B2B company. Understanding that IoT equipment is typ-
ically deployed for many years, the vendors lock themselves
into the IoT market which brings them into a strong market
position. Indeed, they will be able to procure the best operator
offer and thus have operators compete for their IoT equipment
2 http://www.neul.com/neul/
 15

deployments. This is a fundamental shift in how the cellular vision of a truly global Internet of Things.
market operates.
Table I summarizes the main 5G KPIs [66], and highlights
F. IoT Data Privacy and Trust which of the available/emerging technologies are able to meet
them. Cellular technologies, and especially 3GPP LTE, are
Above business opportunities will not emerge if some fun-
among the most appealing technologies in the modern IoT
damental issues around IoT data privacy and trust are provided.
connectivity landscape. They offer wide coverage, relatively
Notably, new trust models are expected to emerge, raising new
low deployment costs, high level of security, access to dedi-
extended requirements for authentication between different
cated spectrum, and simplicity of management. With the MTC
actors, accountability and non-repudiation. In 5G scenarios,
optimized Rel-12/13 and the introduction of NB-IoT, LTE is
several different devices will be interconnected together, each
also both low cost (communication module cost sub 10 USD)
of them with its own security requirements. Thus, for a device
and low power (10 year life time).
to not represent a potential attack for the whole network, it will
The global coverage, along with solid Radio Resource
not be enough to be compliant to the standard it implements.
Management (RRM) algorithms, yields a robustness and relia-
On top of this, very-low cost devices will need to be protected
bility not offered by any competing technologies. The already
using lighter-solutions which do not impact too much on their
deployed infrastructure, which is in essence subsidized through
lifetime. Obviously the current trust model does not fulfill the
data/voice traffic, does not require the deployment of an
needs of the evolved business and technological 5G scenario.
additional IoT infrastructure, such as observed for Zigbee or
Therefore, it should be re-designed, identifying the crucial
LPWA. This, in turn, lowers the deployment barrier of entry
shortcoming, and proposing new suitable solutions [63].
and the running costs to the point that total cost of ownership
Together with faster and more efficient techniques for han-
(TCO) is one of the lowest when compared to competing
dling security procedures (especially needed for low-latency
solutions (despite the higher modem costs and data plans).
scenarios), the 5G security architecture will have to pay
One of the most interesting features of MTC is the ability
particular attention to protection of personal data [64]. The
to offer SLAs even for the most critical and demanding
increased privacy concern has already emerged in society and
industry applications. Such agreements can be honored since
been discussed by the European Commission, standardization
the spectrum is exclusively owned by the operators, in contrast
bodies, including 3GPP and IETF, and many other forums.
to the ISM band used by Zigbee/LP-Wifi/LPWAs. The result
Actually, since 2G, user privacy has been carefully considered.
is that Industrial IoT companies, such as Worldsensing 3 ,
But, till now the advantages offered by the International
can focus on attracting new customers rather than spending
Mobile Subscriber Identity (IMSI) protection have not ap-
resources on solving (and being liable for) connectivity . This
peared to outweigh the complexity of its implementation.
is a very important business proposition which may prove
Thus, solutions for location and identity privacy should be
decisive in the battle for market share.
improved, with respect to those currently used for 4G, trying
Despite the convincing advantages, some serious challenges
to minimize the overhead they introduce.
remain to make MTC an underlying connectivity backbone for
Recently, some important vulnerabilities have been dis-
the Internet of Things. These challenges are pertinent at device
closed in current 4G LTE systems [65]. They are based on the
and networking levels, and when it comes to a viable business
inevitable tradeoff between performance and security which
proposition.
any system must made. 5G puts together and integrates a
From a device point of view, the two largest issues to date
heterogeneous set of wireless access technologies, enabling
(2015) are energy consumption and modem cost. Regarding
seamless connectivity. This might result into an increase of
the energy consumption, given that transmission powers can-
the potential set of vulnerabilities. For data traversing several
not be reduced significantly if communication ranges are to be
different connectivity technologies, vulnerabilities on one of
maintained, the only solution is to facilitate MTC data trans-
them could be exploited to gain access and attack the whole
mission and reception in the shortest time possible. At device
system.
level, this requires a novel approach to device duty cycling
(i.e. the ability to put the MTC device in an ultra-low power
VII. C ONCLUSIONS state). Regarding modem cost, the performance requirements
5G technologies and the Internet of Things are among the of current MTC devices are unnecessarily high for many low
main elements which will shape the future of the Internet end IoT applications. Current 3GPP efforts thus concentrate on
in the coming years. In this paper, we have analyzed in reducing the performance requirements and complexity which
detail the potential of 3GPP-defined 5G technologies for will enable simplified device implementations and lowered
the IoT, by placing them in the context of the current cost.
connectivity landscape for IoT. Differently from previous From a network point of view, the biggest challenge pertains
cellular technologies which were designed essentially for in facilitating the data transfer from and to the devices as
broadband, the requirements which the future 5G networks quickly as possible. Currently, delays occur in radio bearer
will have to satisfy, and particularly those for MTC make 5G establishment and due to congestion in the wireless channel.
communications a particularly good fit for IoT applications. Therefore, entirely novel approaches in radio-bearer establish-
By offering lower cost, lower energy consumption and support
for very large number of devices, 5G is ready to enable the 3 http://www.worldsensing.com/
 16

TABLE I
I OT K EY P ERFORMANCE I NDICATORS (KPI S ) COVERED BY MODERN CONNECTIVITY TECHNOLOGIES .

ZigBee BLE LP-Wifi LPWA 3GPP Rel8 LTE Rel13 & NB-IoT
Scalability 7 7 3 7 3 3
Reliability 7 3 3 7 3 3
Low Power 3 3 3 3 7 3
Low Latency 7 3 3 7 3 3
Large Coverage 7 7 3 3 3 3
Low module cost 3 3 3 3 7 3
Mobility support 7 7 7 7 3 3
Roaming support 7 7 7 7 3 3
SLA support 7 7 7 7 3 3

ment for very short and infrequent IoT data is required in
5G. Furthermore, if scalability is to be ensured, novel random
access protocols need to be designed which are able to cope
with both IoT as well as voice/data traffic.
Our analysis puts in evidence that such a synergy and mutual
shaping between 5G and IoT might also have interesting
implications from the point of view of business models.
Specifically, we have identified the role which 5G communica-
tions might play in the shift of IoT from infrastructure-driven
to business-driven, and we have given some indications on
how the cellular value chain might get transformed by massive
IoT deployments. Our analysis concludes that, once market
demand for IoT services will be created, 5G will constitute an
essential enabler of a full IoT roll-out.
ACKNOWLEDGMENT
This work was partially funded by the European Commis-
sion H2020 5GPPP projects NORMA, VirtuWind, and by
the FANTASTIC-5G project, which receives funding from
the European Union’s Horizon 2020 research and innova-
tion programme under the grant agreement ICT-671660. It
was also partially supported by the CoSDN project, IN-
TER/POLLUX/12/4434480, funded by the Fonds National de
la Recherche Luxembourg. The authors would also like to
thank Prakash Bhat, Vodafone, UK, and Sama Thareefeh for
their valuable input.
R EFERENCES
[1] M. Weiser, “The computer for the 21st century,” Scientific american,
vol. 265, no. 3, pp. 94–104, 1991.
[2] J. Pontin, “Bill joy’s six webs,” in MIT Technology Review, September
2005.
[3] C. Gomez, J. Oller, and J. Paradells, “Overview and evaluation of
bluetooth low energy: An emerging low-power wireless technology,”
Sensors, vol. 12, no. 9, pp. 11 734–11 753, 2012.
[4] M. Siekkinen, M. Hiienkari, J. K. Nurminen, and J. Nieminen, “How
low energy is bluetooth low energy? comparative measurements with
zigbee/802.15.4,” in Wireless Communications and Networking Confer-
ence Workshops (WCNCW), 2012 IEEE. IEEE, 2012, pp. 232–237.
[5] M. Research, “The need for low cost, high reach, wide area connectivity
for the internet of things. a mobile network operator’s perspective.” white
paper, 2014.
[6] D. Bandyopadhyay and J. Sen, “Internet of things: Applications
and challenges in technology and standardization,” Wireless Personal
Communications, vol. 58, no. 1, pp. 49–69, 2011. [Online]. Available:
http://dx.doi.org/10.1007/s11277-011-0288-5
[7] M. R. Palattella, P. Thubert, X. Vilajosana, T. Watteyne, Q. Wang, and
T. Engel, “6tisch wireless industrial networks: Determinism meets ipv6,”
in Internet of Things. Springer, 2014, pp. 111–141.
[8] L. Da Xu, W. He, and S. Li, “Internet of things in industries: A survey,”
Industrial Informatics, IEEE Transactions on, vol. 10, no. 4, pp. 2233–
2243, 2014.
[9] M. Swan, “The quantified self: Fundamental disruption in big data
science and biological discovery,” Big Data, vol. 1, no. 2, pp. 85–99,
2013.
[10] S. Sicari, A. Rizzardi, L. Grieco, and A. Coen-Porisini, “Security,
privacy and trust in internet of things: The road ahead,” Computer
Networks, vol. 76, pp. 146 – 164, 2015. [Online]. Available:
http://www.sciencedirect.com/science/article/pii/S1389128614003971
[11] 3GPP, “Study on Provision of Low-Cost Machine-Type Communica-
tions (MTC) User Equipments (UEs) Based on LTE,” June 2013.
[12] IEEE std. 802.15.4, Part. 15.4: Wireless Medium Access Control (MAC)
and Physical Layer (PHY) Specifications for Low-Rate Wireless Personal
Area Networks (LR-WPANs), IEEE standard for Information Technology,
Sep. 2009.
[13] IETF WG, “IPv6 over Low power Wireless Personal Area Networks
(6LoWPAN).” Available online: http://tools.ietf.org/wg/6lowpan/.
[14] ——, “Routing Over Low Power and Lossy Networks (ROLL).” Avail-
able online: http://tools.ietf.org/wg/roll/.
[15] S. Andreevy, O. Galinina, A. Pyattaev, M. Gerasimenko, T. Tirronen,
J. Torsner, J. Sachs, M. Dohler, and Y. Koucheryavy, “Understanding
the IoT connectivity landscape: A contemporary m2m radio technology
roadmap,” Communications Magazine, IEEE, to appear.
[16] M. R. Palattella, N. Accettura, X. Vilajosana, T. Watteyne, L. A. Grieco,
G. Boggia, and M. Dohler, “Standardized protocol stack for the internet
of (important) things,” Communications Surveys & Tutorials, IEEE,
vol. 15, no. 3, pp. 1389–1406, 2013.
[17] IEEE std. 802.15.4e, Part. 15.4: Low-Rate Wireless Personal Area
Networks (LR-WPANs) Amendament 1: MAC sublayer, IEEE standard
for Information Technology, April 2012.
[18] K. Pister and L. Doherty, “Tsmp: Time synchronized mesh protocol,”
in IASTED International Symposium on Distributed Sensor Networks
(DSN08). Orlando, Florida, USA: IEEE, Nov. 2008.
[19] Industrial Communication Networks - Wireless Communication Net-
work and Communication Profiles, “WirelessHART - IEC 62591.”
Available online: http://www.hartcomm.org, 2010.
[20] ISA/IEC, “ISA100.11a, Wireless System for Au-
tomation, IEC 62734.” Available online:
http://www.isa100wci.org/enUS/Documents/PDF/3405-ISA100-
WirelessSystems-Future-brochWEB-ETSI.aspx, 2011.
[21] IETF WG, “IPv6 over the TSCH mode of IEEE 802.15.4e (6TiSCH).”
Available online: http://tools.ietf.org/wg/6tisch/.
[22] The Bluetooth Special Interest Group, “Specification of the Bluetooth
System, Covered Core Package,” 2010.
[23] J. Nieminen, T. Savolainen, M. Isomaki, B. Patil, Z. Shelby, and
C. Gomez, “Ipv6 over bluetooth(r) low energy,” Working Draft,
IETF Secretariat, Internet-Draft draft-ietf-6lo-btle-17, August 2015,
http://www.ietf.org/internet-drafts/draft-ietf-6lo-btle-17.txt. [Online].
Available: http://www.ietf.org/internet-drafts/draft-ietf-6lo-btle-17.txt
[24] R. Want, “The physical web,” in Proceedings of the 2015 Workshop
on IoT Challenges in Mobile and Industrial Systems, ser. IoT-Sys
’15. New York, NY, USA: ACM, 2015, pp. 1–1. [Online]. Available:
http://doi.acm.org/10.1145/2753476.2753496
[25] E. Khorov, A. Lyakhov, A. Krotov, and A. Guschin, “A survey on
ieee 802.11 ah: An enabling networking technology for smart cities,”
Computer Communications, vol. 58, pp. 53–69, 2015.
[26] T. Adame, A. Bel, B. Bellalta, J. Barcelo, and M. Oliver, “Ieee 802.11 ah:
 17

the wifi approach for m2m communications,” Wireless Communications, [50] M. Mendonca, K. Obraczka, and T. Turletti, “The case for software-
IEEE, vol. 21, no. 6, pp. 144–152, 2014. defined networking in heterogeneous networked environments,” in
[27] T. Rebbeck, M. Mackenzie, and N. Afonso, “Low-powered wireless Proceedings of the 2012 ACM Conference on CoNEXT Student
solutions have the potential to increase the m2m market by over 3 billion Workshop, ser. CoNEXT Student ’12. New York, NY, USA:
connections,” report, September 2014. ACM, 2012, pp. 59–60. [Online]. Available: http://doi.acm.org/10.1145/
[28] ETSI GS LTN 001, “Low Throughput Networks (LTN): Use Cases, 2413247.2413283
Functional Architecture, an Protocols,” September 2014. [51] M. Amani, T. Mahmoodi, M. Tatipamula, and H. Aghvami, “Sdn-
[29] 3GPP TR 33.860, “Study on Security aspect of cellular systems with based data offloading for 5g mobile networks,” ZTE communications
support for ultra low complexity and low throughput Internet of Things,” Magazine, vol. 2, July 2014.
Sept 2015. [52] “Cloud ran,” Ericsson White Paper, Sept 2015.
[30] 3GPP TR 33.863, “Study on battery efficient security for very low [53] D. Boswarthick et al, M2M Communications: A Systems Approach.
throughput Machine Type Communication Devices,” Sept 2015. Wiley Publishing, 2012.
[31] J. Andrews, “Seven ways that hetnets are a cellular paradigm shift,” [54] L. Richardson and S. Ruby, RESTful web services. ” O’Reilly Media,
Communications Magazine, IEEE, vol. 51, no. 3, pp. 136–144, March Inc.”, 2008.
2013. [55] M. B. Alaya, Y. Banouar, T. Monteil, C. Chassot, and K. Drira,
[32] H. Ali-Ahmad, C. Cicconetti, A. de la Oliva, M. Drxler, R. Gupta, “Om2m: Extensible etsi-compliant {M2M} service platform with self-
V. Mancuso, L. Roullet, and V. Sciancalepore, “Crowd: An sdn approach configuration capability,” Procedia Computer Science, vol. 32, pp. 1079
for densenets,” in Proceedings of the 2013 Second European Workshop – 1086, 2014.
on Software Defined Networks, ser. EWSDN ’13. Washington, DC, [56] ETSI TS 102 690, “Machine-to-Machine communications (M2M); Func-
USA: IEEE Computer Society, 2013, pp. 25–31. [Online]. Available: tional architecture,” October 2013.
http://dx.doi.org/10.1109/EWSDN.2013.11 [57] L. Grieco, M. Ben Alaya, T. Monteil, and K. Drira, “Architecting
[33] B. Soret, K. Pedersen, N. Jorgensen, and V. Fernandez-Lopez, “In- information centric etsi-m2m systems,” in Pervasive Computing and
terference coordination for dense wireless networks,” Communications Communications Workshops (PERCOM Workshops), 2014 IEEE Inter-
Magazine, IEEE, vol. 53, no. 1, pp. 102–109, January 2015. national Conference on, March 2014, pp. 211–214.
[34] F. Boccardi, R. Heath, A. Lozano, T. Marzetta, and P. Popovski, “Five [58] J. Swetina, G. Lu, P. Jacobs, F. Ennesser, and J. Song, “Toward
disruptive technology directions for 5g,” Communications Magazine, a standardized common m2m service layer platform: Introduction to
IEEE, vol. 52, no. 2, pp. 74–80, February 2014. onem2m,” IEEE Wireless Communications, vol. 21, no. 3, pp. 20–26,
[35] M. Condoluci, M. Dohler, G. Araniti, A. Molinaro, and K. Zheng, June 2014.
“Toward 5g densenets: architectural advances for effective machine-type [59] TS-0003.V1.0.1, “Security Solutions,” January 2015.
communications over femtocells,” Communications Magazine, IEEE, [60] (2015) GE M2M study. [Online]. Available:
vol. 53, no. 1, pp. 134–141, January 2015. http://www.forbes.com/sites/greatspeculations/2014/11/12/
[36] S. Chen and J. Zhao, “The requirements, challenges, and technologies ge-is-beginning-to-see-strong-returns-on-its-industrial-internet-investments/
for 5g of terrestrial mobile telecommunication,” IEEE Communications [61] A. Frost, “Smart city as a service,” white paper, 2015.
Magazine, May 2014. [62] S. Thareefeh, “Zero-capex smart city,” MSc Thesis, King’s College
[37] X. Zhang, X. Shen, and L.-L. Xie, “Joint subcarrier and power allocation London, Tech. Rep., September 2015, supervised by Prof. M. Dohler.
for cooperative communications in lte-advanced networks,” Wireless [63] Ericsson, “5g security, scenarios and solutions,” white paper, June 2015.
Communications, IEEE Transactions on, vol. 13, no. 2, pp. 658–668, [64] G. M. Koien and V. A. Oleshchuk, Aspects of Personal Privacy in
February 2014. Communications-Problems, Technology and Solutions. River Publish-
[38] N. Zlatanov, A. Ikhlef, T. Islam, and R. Schober, “Buffer-aided cooper- ers, 2013.
ative communications: opportunities and challenges,” Communications [65] A. Shaik, R. Borgaonkar, N. Asokan, V. Niemi, and J.-P. Seifert,
Magazine, IEEE, vol. 52, no. 4, pp. 146–153, April 2014. “Practical attacks against privacy and availability in 4g/lte mobile
[39] X. Tao, X. Xu, and Q. Cui, “An overview of cooperative communica- communication systems,” arXiv preprint arXiv:1510.07563, 2015.
tions,” Communications Magazine, IEEE, vol. 50, no. 6, pp. 65–71, June [66] NGMN Alliance, “5G White Paper - Executive Version,” NGMN White
2012. Paper, Dec 2014.
[40] M. Tehrani, M. Uysal, and H. Yanikomeroglu, “Device-to-device com-
munication in 5g cellular networks: challenges, solutions, and future
directions,” Communications Magazine, IEEE, vol. 52, no. 5, pp. 86–
92, May 2014.
[41] H. Elshaer, F. Boccardi, M. Dohler, and R. Irmer, “Downlink and uplink
decoupling: a disruptive architectural design for 5g networks,” in Global
Communications Conference (GLOBECOM), 2014 IEEE. IEEE, 2014,
pp. 1798–1803.
[42] ——, “Load & backhaul aware decoupled downlink/uplink access in 5g
systems,” arXiv preprint arXiv:1410.6680, 2014.
[43] F. Boccardi, J. Andrews, H. Elshaer, M. Dohler, S. Parkvall, P. Popovski,
and S. Singh, “Why to decouple the uplink and downlink in cellular
networks and how to do it,” arXiv preprint arXiv:1503.06746, 2015.
[44] P. Rost, C. Bernardos, A. Domenico, M. Girolamo, M. Lalam,
A. Maeder, D. Sabella, and D. Wbben, “Cloud technologies for flexible
5g radio access networks,” Communications Magazine, IEEE, vol. 52,
no. 5, pp. 68–76, May 2014.
[45] H. Lateef, A. Imran, M. Imran, L. Giupponi, and M. Dohler, “Lte-
advanced self-organising network conflicts and coordination algorithms,”
IEEE Wireless Communication Magazine, in press.
[46] F. Granelli, A. Gebremariam, M. Usman, F. Cugini, V. Stamati, M. Al-
itska, and P. Chatzimisios, “Software defined and virtualized wireless
access in future wireless networks: scenarios and standards,” Communi-
cations Magazine, IEEE, vol. 53, no. 6, pp. 26–34, June 2015.
[47] H.-H. Cho, C.-F. Lai, T. K. Shih, and H.-C. Chao, “Integration of sdr
and sdn for 5g,” Access, IEEE, vol. 2, pp. 1196–1204, 2014.
[48] A. Hakiri and P. Berthou, “Leveraging SDN for the 5g networks:
Trends, prospects and challenges,” CoRR, vol. abs/1506.02876, 2015.
[Online]. Available: http://arxiv.org/abs/1506.02876
[49] D. Kreutz, F. M. Ramos, P. Esteves Verissimo, C. Esteve Rothenberg,
S. Azodolmolky, and S. Uhlig, “Software-defined networking: A com-
prehensive survey,” proceedings of the IEEE, vol. 103, no. 1, pp. 14–76,
2015.